The work practices used and the design of the machinery
The work practices was so bad, they never inspected the brake rod and trunnion during the 21 years. The design of machinery was so bad too. The brake rod was always in tension with the brake on and brake off. And the emergency stop system was effective when the brake rod was effective. If the brake rod is broken, the system can not do anything to stop the accident. In the cross-head trunnion, the top surfaces of the axle in contact with the bearing pads were fretted and scored, they were badly worn. Note that they were made of same material which was mild steel and it took a lot of friction. All these design was so poor and dangerous.
The causes of the accident and the specific failure mechanisms of the components.
We can get answer from the analysis. We know, from the chemical analysis, mechanical testing, the steel meets the BS 970:1947, but has a very low Charpy energy by modern standards and the microstructure indicates a normalized steel. It means the steel is a eligible material. So there is no problem.
Now we enter the realm of the mechanical or structural engineer and consider, first, the loading conditions. Axial and Bending Loads—The broken brake rod replaced with a new one provided the investigators with a working system on which to measure the loads. Four strain gauges were attached to the rod at a distance of 533 mm from the lower end with their active direction parallel to the long axis. The gages, numbered 1 to 4, were located 90○ apart around the circumference of the rod. The result, converted to stress, are displayed by the gages. Dealing with the tensile stress first; this changes from an average value of 66 MN m-2 with the brake on to 73 MN m-2 with the brake off. This is precisely the result which would be expected from the kinematics of the braking system: slightly more tension with the brake off. What is surprising is the presence of large bending stresses. Indeed, the presence of the trunnion in the design suggests that the brake rod was not intended to bend when the main level rotated during braking. The major bending stresses are in the (2-4) plane and these show large changes when the brake is moved form on to off. There are also bending stresses in the (1-3) plane, but these don’t change so much. It was found that the mild steel bearing pad surfaces on the main lever were badly worn. Similarly in the crosshead, the top surfaces of the steel axle in contact with the bearing pads were fretted and scored. The importance of the lack of lubrication of the trunnion was clearly demonstrated by lubricating the crosshead trunnion axle bearing surfaces with graphite grease and then measuring the turnnion rotation relative to the main lever and the associated stress as a function of the number of on-off brake cycles. After only 13 brake cycles the alternating stress had increased to more than ±75 MN. It appears that high bearing pressures caused the lubricant to be progressively squeezed out of the gap between the trunnion axle and the bearing pads. We can calculate the most highly stressed case occurs with the brake in the off position, when a bending stress of 103 MN m-2 m and a tensile stress of MN m-2 produce a combined stress of MN m-2.
Stresses at the fracture site. Strains could not be measured at the fracture site because of the threads. Thus, stresses were determined from strain measurements taken at the nearest accessible position. To relate the bending stress determined at the position of the strain gages to that at the fracture site it is necessary to model the loading of the brake rod. The simplest model assumes that, due to uneven loading on the nut, a moment is introduced at the top of the rod. The moment will remain constant along the length of the rod until it is reacted by the nut bearing against the collar beneath the main lever. This model predicts a maximum bending stress (that is, with the brake off) at the fracture site of 157 MN m-2 and axial stress of 97 MN m-2 The combined tensile stress is therefore 254 MN m-2
Based on yield stress, the factor of safety is 1.36, and on tensile strength, it is 2.36. Thus, the combined static effect of bending and tension is not sufficient to cause an overload failure of the uncracked brake rod. The stresses concentration factor at the thread roots is about 4. Taking this into account, there will be some localized plastic deformation at the thread roots.
Fracture toughness of the Brake Rod Steel—Fracture toughness was determined on standard, notched, three-point bend specimens machined from the brake rod. By cyclic loading, a fatigue crack was formed at the notch and the load require to make this propagate was measured. As these values of KQ are the only ones we have and the fracture surfaces do not show extensive plastic deformation, I will take the average, approximately 44.5 MN m-3/2, as being KIC for the material. We can calculate the σ=K/Y(πa)1/2 andσ=49 MN m-2. Thus, the rod with a 27mm crack needs a stress of only 49 MN m-2 to initiate brittle fracture. This should be compared with the estimated stress at the fracture site, made from strain gage measurements on an uncracked rod, of 97 MN m-2. It is clear that all the load on the cracked rod was not reaching the fracture site. The only thing between the strain gage site and the fracture site is the crosshead trunnion axle and it is here that some of the load on the rod must have been reacted. Indeed, the investigators observed rubbing marks on the brake rod where it passed through the trunnion, Contact between rod and trunnion would be associated with a frictional force which would prevent some of the axial load from being transmitted to the lower end of the rod. This may always have been the case, if the rod was assembled to one side of the hole in the trunnion or if the elastic deflection was enough to cause contact. Alternatively, the crack may have caused the rod to become so flexible as to bring one side or both sides of the rod into contact with the sides of the trunnion block. So now let’s do the fatigue analysis. We have to estimate the change in bending stress. The bending stress is given byσ=σBsinθ+σccosθ.
Substituting in the values of σB and σc for the brake off and brake on conditions. The changes which simplifies to Δσ=19sinθ+132cosθ. The max value ofΔσ occurs when the value of θ is θm, so theθm=8.2〇 and theΔσm=133 MN. This result is significant in that it clears up an anomaly from the first analysis. The angular position of the maximum bending stress varied from θ=-122.5 with the brake on to –20.4 with the brake off and yet the fracture origin was at roughly 0. the maximum change in bending stress, being at 8.2, is much more consistent with a fracture origin at 0.
To check the material response to the stresses in the threaded section we need to calculate the mean stress and the stress amplitude and plot these points on a Goodman diagram. The mean stress level at the position of maximum stress amplitude has two components: the contribution of bending and the contribution of the axial stress. The bending stresses at the position of the maximum stress amplitude are at θ=8.2. the mean of the value is 24.5 MN m-2, the value of the mean bending stress in the threaded portion of the rod is 37 MN m-2, as the axial stress is uniform over the section, its mean value is 92 MN m-2 which is simply given by the average of the axial stresses with the brake on and brake off. The mean stress at the position of maximum stress amplitude is then simply given by the sum of the axial and bending components which is 92+37=129 MN m-2 The stress amplitude is again given by a contribution from bending and axial stresses. The change in stress due to bending was calculated to be 202. In additional there is a change in stress due to axial loading, of 9 MN m-2. The sum of these two contributions gives the total change in stress as 202+9=211 MN m-2. The stress amplitude is just half this value. We conclude that the cyclic stress of maximum amplitude is that shown graphically . So the Goodman diagram predicts that fatigue will occur in bending. This result is suggests that when the rod was new and the bearing seized then all the load above the trunnion axle was transmitted to the fracture site and, it was only after the crack had grown, probably to a significant length, that the rod was able to flex sufficiently to touch the trunnion block, thereby reducing the load on the fracture site. From the estimating lifetime analysis, it shows there were 7x105 cycles the broke rod experience in its service lifetime of 21 years and it took roughly 98000 cycles to grow a crack from 0.75mm to 3mm long. The estimates of lifetime are incomplete. However, they fully support the verdict of the investigation: that the brake rod became progressively cracked by fatigue until finally the fatigue crack became unstable causing brittle fracture. The agent of fatigue was an unforeseen bending which arose because the trunnion mechanism was ineffective.
The center rod in the spring nest is an example of a single-line component, in that the operation of the mechanical brakes on the winding engine depended completely on it. This type of system was taken out of service following the accident. the official report states that single-line components should be either eliminated or so designed as to prevent danger.
So, from the analysis above, we can identify fatigue as the cause of this failure(ineffective trunnion bearing) and to highlight bending moments resulting from friction at the trunnion as the loading which caused this fatigue crack growth using straightforward techniques. The analysis presented here has taken this understanding further and identified a variation in principal bending axis, between the brake on and brake off positions. This produces a maximum change in bending stress—the critical fatigue loading—more in line with the fracture origin. At a more general level the case study illustrates the danger of single line components. The brake rod had lasted for 21years despite being made from what, by modern standards, would be called a dirty steel and having a cut as opposed to a rolled thread.
Some recommendations would prevent the disaster from happening again.
I think we need to make a second independent brake system to prevent the accident, so if the first system can not be effective, the second one can be used.
Nobody inspected the brake rod during 21 years is a terrible thing, we need regular inspection of the brake rod and trunnion, and if it is necessary, we should change a new one to replace the old. It can reduce a lot of friction that the inset bearing pads and the trunnion axles are made of different material and we add some balls and lube between them.